ChatGPT x Awareways: opportunities & challenges for Learning & Development

Starting from backgrounds in psychology, education, UX/UI and gamification, the L&D team is always looking to improve our security awareness programs. Looking for what needs to be explored to continue to facilitate sustainable behavioral change.

When the task is providing challenging new interventions in our learning experience platform, always focusing on maximal learning efficiency, how can an AI-tool like ChatGPT support that mission?

ChatGPT as a sounding board

“The first thing that comes to mind”, says Pascal Koelemij, “is that ChatGPT generates a wealth of inspiration really quickly and in an uncomplicated manner. If you want to address a particular question, or cover a particular subject, AI can be super helpful.”

“Another is that if ChatGPT can answer a question given the relevant context, it is a good first test whether or not a user on our security awareness platform will be able to do the same. So I’d say we first and foremost use it as a sounding board. That’s also where I see its current limitations. Because as interesting as it is, we’re not quite near its full potential.”

'When you talk about content production, ChatGPT provides semi-finished products of best.'

Cookie cutter content

“In general, the approach to get useful input out of ChatGPT is via one of two ways. It’s either ‘human generated, AI validated’ or the other way around, ‘AI generated, human validated’. The earlier example of sound-boarding is a demonstration of the former: you draft a piece of content and use AI to assess quality and usability. The other route would be to write a prompt for ChatGPT, asking it to come up with a piece of content.”

“In terms of content production however, ChatGPT is – at best – good in providing semi-finished pieces. What’s more, there’s a huge potential for [AI] output to turn all online content into cookie cutter monotony, with the commonality of lacking any originality or distinction.” We’re witnessing a similar trend in the current (redesign) of corporate logos, all using the same minimalist approach, with brands morphing into ‘blands’.

Creativity

“That’s because at its core, it lacks creativity. If we ask the program to come up with a creative new name for one of our skill games, it equates ‘creative’ with ‘use alliteration’, and offers a list of dozens of completely useless suggestions. Even if your next prompt is ‘don’t’ use alliteration’, it will still come back with ‘amazing awareness adventure’. Which is to say, it simply cannot replace – or even match – human creativity. The trick to get the best use out of it then, is to approach the output as just that: semi-finished.”

“So, if you ask it to generate some ideas, it can provide a jumping-off point, sure. But if you ask for actual, usable output, it at best offers a step in the right direction. Which is to say, it can give you an idea of what it is you are not looking for. Which is helpful, in a way, but hardly an argument to replace your in-house copywriter. In fact, brands can retain a certain competitive advantage down the line by maintaining and cultivating their in-house creativity, especially if others decide to outsource to AI.”

“Of course, we shouldn’t ignore the potential either. Don’t overestimate, but do not underestimate either. Down the line, there’ll be a so-called ‘plateau of productivity’ where the tool/innovation is embedded in daily work routines. For now, it’s good to be critical, use the tool for what it is, and keep an eye out for new developments. And: make sure that you have solid ground-rules for AI use as well. We do.”

…

'It simply cannot replace or even match human creativity.'

'Treat AI tools like a knowledgeable but overconfident friend.'

Awareways x ChatGPT: security measures

As AI is becoming more and more common place and legislative safeguards are being sought, we’ve set up some ground rules for its applications at Awareways as well. Use of AI is by no means prohibited and there is a ChatGPT-4 account available, but as with all our endeavors, information security is a top priority. In other words, we embrace technological developments, but have internal agreements to deal with them responsibly.

To make sure everyone knows how that relates to the use of AI, we’ve established a list of do’s and don’ts at the office. A dynamic and ever-evolving list, as AI keeps developing fast.

The list includes DO’s like ‘use it to create new insights’ and ‘be aware of the limitations of the system and its potential for error’. “Treat AI tools like a knowledgeable but overconfident friend” is the important message there.
DON’TS include ‘do not enter sensitive data’ (whether internal, personal or client-related), and ‘do not immediately take the answers as true’.

In conclusion

“In the end, it’s important to take a step back and understand that we should not expect magic from a tool that relies on knowledge and context that’s already there”, says Pascal. “ChatGPT’s source for everything is the internet as a whole – up to its current limitations – as a source of knowledge that already exists, that’s already availabe. So, it’s not going to come up with anything ground-breaking, because it’s not working with anything ground-breaking.”

“More importantly, if we relate that to cybersecurity and more specifically security awareness, there’s never going to be a final station. We’ll never reach a point where nobody will ever click on a bad link in an email ever again. ‘We’ve made it, we’re at 100 percent security’ is simply not a realistic destination.”

“The maturity of any organization, any security culture, has no ceiling. The most mature you can get is being ready for all current threats and challenges while preparing for challenges around the corner, or at the very least acknowledging that there will be new challenges. Awareness is a continuous effort, where change is the only constant.”